The year-to-year snow fraction link to runoff ratio in 268 US snow basins followed the winter share of precipitation, not the rain or snow phase, in water years 1981-2014

Berghuijs et al. (2014) reported that a higher snow fraction of precipitation (fs) gives more runoff, across catchments and from year to year. From year to year, fs changes with temperature and with the season of precipitation, which also change evaporation. We used 268 CAMELS reference basins with a long-term snow fraction of 0.15 or more and 8,895 basin water years (1981-2014). We split fs into a monthly part (from monthly precipitation and the daily temperatures of each month) and a within-month part (fs_w), which changes only when precipitation falls on colder or warmer days of the same month. In each basin we fitted the annual runoff ratio (Q/P) and pooled the slopes with a bootstrap that resamples whole water years. Per +0.1 fs, the plain link was +0.018 (95% interval +0.013 to +0.025). With temperature and the October-March share of precipitation as controls, it was -0.002 (-0.010 to +0.007). The primary fs_w slope was -0.005 (-0.017 to +0.007). As a positive control, fs_w moved flow into April-July (+0.021 share per +0.1, interval +0.002 to +0.040), 78% of the shift from the full fs. Three snow rules, a storm-size control, subsets and a next-year placebo gave the same null result. Thus the year-to-year link in these data follows the season of precipitation, not the phase itself. Limitations: within-month snow is often a short-term store, so the test does not measure the effect of a long-term loss of snowpack. Other limits are a temperature phase rule on basin means, snow gauge errors, 1981-2014 data and annual totals only.

snow fraction · rain-snow partitioning · runoff ratio · water availability · CAMELS · Daymet · within-catchment variability · precipitation seasonality · Budyko · positive control · null result